My Perl QA Hackathon 2016

By BooK on April 30, 2016 11:05 PM

I attended the Perl QA Hackathon 2016 in Rugby. I didn't come with any specific projects, and planned to participate in some of the discussions, help others if possible, learn a bit of Perl 6 and if all else failed, work on some of my own CPAN projects. I did a little bit of all.

Pair programming

I spent a lot of time with Olivier Mengué (the other French participant):

We worked together on Dist::Zilla::Role::PrereqsScanner, a refactoring of Dist::Zilla::Plugin::AutoPrereqs
We talked a lot about Git::Database, my next Git-related module, and I went through various changes of architecture with it. I'ts still in a state of flux, but I'm hoping for a first release soon.

I also paired a little with Liz, see below.

Meetings and discussions

I didn't attend all meetings, only those that I thought I could either contribute to or learn from:

A presentation of Pakket by Sawyer X
The CPAN River discussions
The discussion on renaming the QA Hackathon so that the world outside stops think its a competitive event where multiple teams try to rush some minimally working code to win a prize. A temporary conclusion to the discussion was that the words "Perl", "reliability", and "infrastructure" were important.

Debian packages of CPAN distributions

After the CPAN River discussions, I somehow found myself wondering about CPAN distributions packaged in Debian. So I worked on producig a complete mapping of CPAN distributions to Debian packages using the UDD (Ultimate Debian Database).

By the end of the hackathon, I had mapped over 3300 CPAN distributions. The next step is a MetaCPAN PR to show this information on a distribution page.

While working on this, I realized that once we know which Debian packages are actually CPAN distributions, then all the Debian packages that depend on those can be counted in the CPAN River metrics. This could be the first step in including DarkPAN (i.e. Perl code not on CPAN) in the CPAN River metrics.

I discussed this topic with Olivier, who decided to work on the extraction of dependencies from any Perl-based project. At some point in the future, we will be able to bring in GreyPAN (Open Source DarkPAN) projects one by one into the light...

Git version numbers in Perl 5 and Perl 6

Before the hackathon, I had spent an evening with ab5tract trying to conjure a Perl 6 version of Git::Version+Git::Version::Compare.

During the hackathon, I showed the project with Liz, and we discussed about the Version type in Perl 6, which led her to push some patches to Rakudo to improve it.

Inspired by the Perl 6 spectest for Version, I improved the test suite for Git::Version::Compare to be much more comprehensive (it jumped from about 400 hundred tests to more than 11000 for the same dataset), and started working on porting this approach as stress tests for the Perl 6 Version. And did a pull request after the hackathon.

BooK